You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Understanding the V5 architecture, and what affect its memory size
First, outlining the status quo, V5 has proven to beat out all existing V4, and V5+wavenet architecture in being able to retain memory for L24/D2048. The attached is the high level performance
The goal, is to figure out how this scales, by layer/embedding size/head size/ etc
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
First, outlining the status quo, V5 has proven to beat out all existing V4, and V5+wavenet architecture in being able to retain memory for L24/D2048. The attached is the high level performance
The goal, is to figure out how this scales, by layer/embedding size/head size/ etc
Follow on the following discord thread: https://discord.com/channels/992359628979568762/1142397705314906182/1142397998853279744
Beta Was this translation helpful? Give feedback.
All reactions